When does the refinement of a paracompact topology remain paracompact? Let $(X,\tau)$ be a Hausdorff paracompact space. Let $\tau'$ be the smallest $P$-topology refining $(X,\tau)$, i.e. the topology which has for base the $G_\delta$-subsets of $(X,\tau)$.
Is it true that $(X,\tau')$ is again paracompact?
If not, does the result hold under extra assumptions (e.g. hereditary paracompactness or even more)?
EDIT: Given the first answer received, I'll add a question:
Is there a Hausdorff paracompact space $(X,\tau)$ where every point has a local base of size $\leq \omega_1$ such that $(X,\tau')$ is not paracompact?
 A: A counterexample to this claim is stated in the paper Normality and paracompactness in box products by Eric K.van Douwen and Kenneth Kunen (1990).
If $\kappa$ is an infinite cardinal, then the $<\kappa$-box topology on a product $\prod_{i\in I}X_i$ of topological spaces is the topology where the basic open sets are of the form $\prod_{i\in I}U_i$ where each $U_i$ is an open subset of $X_i$ and where $|\{i\in I\mid U_i\neq X_i\}|<\kappa$. The $P$-space coreflection of a space $(X,\tau)$ is the space $(X,\tau')$ where $\tau'$ is generated by the $G_\delta$ subsets of $(X,\tau)$.
Theorem: If $\kappa$ is an infinite cardinal, $\kappa$ is given the discrete topology, and $\kappa^{(\kappa^+)}$ is given the $<\kappa$-box topology, then
$\kappa^{(\kappa^+)}$ is not normal.
In particular, if we assume the continuum hypothesis, then
$\mathfrak{c}^{(\mathfrak{c}^+)}$ is not normal, but $\mathfrak{c}^{(\mathfrak{c}^+)}$ is the $P$-space coreflection of the compact space $[0,1]^{\mathfrak{c}^+}$ with the product topology.
